Re: Intolerable CVS behaviour

[Miguel de Icaza <miguel ximian com>]

> I do find it weird though that we can't even trust the people who we
> give CVS access anymore to play by the rules. Doesn't this indicate
> a bigger/other problem?

It is not a matter of trust, it is a matter of risk.

We are humans, and we make mistakes.  This means, that we might do
something wrong that we did not intend to happen.  We might have the
best intentions, but still can make mistakes.

Now, if every week there is a 0.1% chance (one in a thousand) to make
a mistake on the CVS and you have one developer, the problem is not
too bad.  If you have a thousand developers, there is one mistake
per-week that will happen.

Lets talk programmer-talk.  Assertions and pre-conditions in Gtk+ are
a good example of a mechanical device that can help programmers. 

It is like passing a NULL pointer to gtk_label_new.  It is not that
you can not trust programmer's to pass the right argument, it is just
that you are being careful and assume that people will make mistakes
and you will be able to help them by the means of a mechanical
procedure (assertions) to compensate for their mistake.

Now, what mechanical devices can we think of that would reduce the
risk of a mistake on the CVS, and hence wasting precious from the
contributors and the people running the repository?

I can think of a few:

	* Translators can be given automated tools that would do the
          "right thing" without them having to become CVS experts.

	* Modules with strict CVS commit rules can be mechanically
          managed, without assuming that everyone will be educated [1]
          on time, or will do the right thing given then chances of
          errors.

[1] Proof that education does not always work: Try to get the media to
understand the difference between `Hackers' and `Crackers'.  There are
many reasons why 

Best wishes,
Miguel.